Uber sees OSM as a valuable tool and uses it in internal models to help determine time + distance estimates for fare calculation and to optimize driver and rider matching. In certain circumstances, we are able to observe that road directionality and turn restrictions are not up-to-date. We plan to explore these issues and, where appropriate, submit updates to correct the issue. We have selected the Delhi NCR region as the first city for this initiative.
We do not plan to make any large-scale, machine-generated edits for this project. All edits will be made by a small team of individuals based in Palo Alto, California, USA. The team will be using the available data sources in the JOSM tool for their corrections and validations. We will share the profiles of our editors on Uber’s OSM page soon. The edits will be made according to the Organized Editing Best Practices and India guidelines. In addition, Uber employees based in Delhi will be able to participate in this project by leveraging their local knowledge to help identify data issues.
For this project, our team will focus on adding and modifying the following OSM features:
This is a great initiative, thanks for the update Uber team.
The team will be using the available data sources in the JOSM tool for their corrections and validations
Can you provide more information on the data sources that are going to be used for this mapping push? All the features mentioned above needs street-level imagery or field survey for confirmation apart from satellite imagery. Are these data sources being derived from your internal data pipeline or your team is going to use any external data sources for mapping these features?
@sunil-kaw, welcome for such efforts. validation process starts from the grond truth. As @rorym recommends, it would be best for all if uber joins the funding programme. Moreover, why NCR delhi ? And how is it selected ? Is there not osm contributors there ? Are uber employees are already part of local osm community. ? Using josm tool for validation does not validates everything. May be geometry could be validated, but still directionality and turn restrictions require huge osm labour. To get such high quality graph parameters that might help with ubers ride optimization task. Does this mean uber might use osm data while finding the issue ? Moreover technically and socially OSM is not just a tool, it is both community and tool that makes osm as what it is. Since osm community AFAIK, has started supporting the contributors financially, does ur employees will get it too ? Its nice to have ubers few open source projects. Since u have said that uber already consumes the osm data, I would like to know what kind of operations are being run on the osm data. ???
@sunil-kaw I appreciate Uber’s contribution to OSM. I have a suggestion for your team. Please do not rely on the aerial imagery available in OSM. It’s outdated and contributors like me use local knowledge to map. Some of your team members are overwriting those changes thinking that aerial imaginary is much more accurate. Kindly ask them to see the update history of a place and contact the local community member if they have any doubt about the area.
I am posting here an image of an edit which makes zero sense but was still added. Kindly educate them about responsible editing and asking for help when needed.
I notice Uber haven’t done that yet, and I would encourage you to prioritize these steps, and pause your mapping until you have demonstrated that you want to be good citizen of the OSM ecosystem.
I have also noticed that some of the user profile pages of the people working for you have a boilerplate text very commonly used by a big outsourcing firm: “I am and I am happy to be contributing to the Uber OSM project. In my free time, ”. While a personal touch in a user profile is certainly nice, seeing hundreds of user profiles following that exact same structure means that it’s not really personal at all. For us, it would be much more useful if instead of a list of hobbies, the profile had concrete links to the projects they are working on, so we can understand what data sources they are using and what their goals are, and whom we can contact if something goes wrong. If you are using workers from an outsourcing company, then it would also be helpful to be transparent about this (because their training will often not have come from the client but from the employer, and if they make systematic mistakes we know that we need to approach the employer to improve the training). Ideally, the user profile would then say: “Hello, I am X, I am employed at and I work on the following projects for Uber: ”.